Decision trees in the identification of Space Models

Abstract

Space Models are frequently used by location-aware applications to identify the location context at which a particular user is. Attending to the identified location, these applications adapt their behaviour in order to provide relevant services to the user. Therefore, Space Models play an important role in the dynamic adaptation of location and context-aware applications. These models can be manually created by analysing specific geographic regions and assigning them the appropriate classes (for instance, rural area, urban area, suburban area, city center, among others). This paper presents an alternative process for the identification of Space Models that is based on the use of Decision Trees. Given the performance achieved by this approach in the classification of the location contexts, the resulting model was used to create a Space Model for the Portuguese territory. This Space Model is now available to be used by location-aware applications.